About

Hadi Meidani is an Associate Professor in the Department of Civil and Environmental Engineering at the University of Illinois at Urbana-Champaign (UIUC). His research focuses on transforming how engineering systems are modeled, designed, and operated by advancing a new paradigm of AI-driven scientific computing. His work includes physics-informed machine learning, neural operators, and graph-based AI models to accelerate traditional simulation and enable scalable digital twins for infrastructure systems, transportation, structural mechanics, and biomedical applications. Dr. Meidani has received recognition such as an NSF CAREER Award for his contributions to fast computational models for infrastructure networks. His team has won awards from data competitions related to railroad engineering, and his research has been sponsored by federal agencies including NSF, DOE, and DOT. Prior to joining UIUC, he held postdoctoral positions at USC and the University of Utah, and he is the Chair of the Machine Learning Committee of the ASCE Engineering Mechanics Institute.

Research topics

• Artificial Intelligence
• Computer Science
• Mathematics
• Machine Learning
• Algorithm
• Applied mathematics
• Mathematical analysis
• Mathematical optimization

Selected publications

• NuGraph2 with context-aware inputs: physics-inspired improvements in semantic segmentation

  Journal of High Energy Physics · 2026-03-16

  articleOpen accessSenior author

  A bstract Graph neural networks have recently shown strong promise for event reconstruction tasks in Liquid Argon Time Projection Chambers, yet their performance remains limited for underrepresented classes of particles, such as Michel electrons. In this work, we investigate physics-informed strategies to improve semantic segmentation within the NuGraph2 architecture. We explore three complementary approaches: (i) enriching the input representation with context-aware features derived from detector geometry and track continuity, (ii) introducing auxiliary decoders to capture class-level correlations, and (iii) incorporating energy-based regularization terms motivated by Michel electron energy distributions. Experiments on MicroBooNE public datasets show that physics-inspired feature augmentation yields the largest gains, particularly boosting Michel electron precision and recall by disentangling overlapping latent space regions. In contrast, auxiliary decoders and energy-regularization terms provided limited improvements, partly due to the hit-level nature of NuGraph2, which lacks explicit particle- or event-level representations. Our findings highlight that embedding physics context directly into node-level inputs is more effective than imposing task-specific auxiliary losses, and suggest that future hierarchical architectures such as NuGraph3, with explicit particle- and event-level reasoning, will provide a more natural setting for advanced decoders and physics-based regularization. The code for this work is publicly available on Github at https://github.com/vitorgrizzi/nugraph_phys/tree/main_phys .

  PublisherOA PDFDOI

• Work in Progress: Assessing AI Integration in School Education

  2026-02-23

  articleOpen access

  Artificial Intelligence (AI) is increasingly influencing the educational landscape, offering new opportunities for personalized learning and administrative efficiency.This study investigates AI integration in secondary education by examining teachers' familiarity, willingness to adopt AIdriven tools, and perceived challenges.A survey was conducted among nine teachers from two high schools, covering subjects such as mathematics, history, and art.Preliminary findings indicate that while teachers acknowledge AI's potential benefits, barriers such as lack of training, privacy concerns, and institutional resistance hinder adoption.Future work includes pilot studies and codesign efforts with educators to develop an AI-based teaching assistant system (ATAS).This study also explores how lessons from secondary education may inform AI adoption in post-secondary settings, particularly in engineering and physics departments, where AI applications are becoming increasingly relevant.

  PublisherOA PDFDOI

• Sequential neural operator transformer for high-fidelity surrogates of time-dependent non-linear partial differential equations

  Engineering Applications of Artificial Intelligence · 2026-03-07 · 1 citations

  articleOpen accessSenior author

  Partial differential equations (PDEs) are fundamental to modeling complex and nonlinear physical phenomena, but their numerical solution often requires significant computational resources, particularly when a large number of forward full solution evaluations are necessary, such as in design, optimization, sensitivity analysis, and uncertainty quantification. Recent advances in artificial intelligence – particularly operator learning – have enabled surrogate models that efficiently predict full-field PDE solutions; however, these models often struggle with accuracy and robustness when faced with highly nonlinear responses driven by sequential input functions. To address these challenges, we propose the Sequential Neural Operator Transformer (S-NOT), an architecture that combines gated recurrent units (GRUs) with the self-attention mechanism of transformers to address time-dependent, nonlinear PDEs. Unlike sequential-deep operator networks(S-DON), which use a dot product to merge encoded outputs from the branch and trunk sub-networks, S-NOT leverages attention to better capture intricate dependencies between sequential inputs and spatial query points. We benchmark S-NOT on three challenging datasets from real-world applications with plastic and thermo-viscoplastic highly nonlinear material responses: multiphysics steel solidification, a three dimensional (3D) lug specimen, and a dogbone specimen under temporal and path-dependent loadings. The results show that S-NOT yields prediction errors up to 4.5 times smaller than S-DON even for data outliers. Furthermore, S-NOT provides an acceleration of 4 orders-of-magnitude compared to traditional finite element method simulations, demonstrating its accuracy and robustness for drastically accelerating computational frameworks in scientific and engineering applications.

  PublisherDOI

• Hybrid Computer Vision Model to Predict Lung Cancer in Diverse Populations

  JCO Clinical Cancer Informatics · 2026-01-01

  article

  PURPOSE: Disparities in lung cancer incidence exist in Black populations, and screening criteria underserve Black populations due to disparately elevated risk in the screening-eligible population. Prediction models that integrate clinical and imaging-based features to individualize lung cancer risk are a potential means to mitigate these disparities. METHODS: This multicenter (National Lung Screening Trial [NLST]) and catchment population-based (University of Illinois Health [UIH], urban and suburban Cook County) cross-sectional study used participants at risk of lung cancer with available lung computed tomography (CT) imaging and follow-up between the years 2015 and 2024. In all, 53,452 in NLST and 11,654 in UIH were included on the basis of age and tobacco use-based risk factors for lung cancer. Cohorts were used for training and testing of deep and machine learning models using clinical features alone or combined with CT image features (hybrid computer vision). RESULTS: An optimized seven-feature clinical model achieved receiver operating characteristic (ROC)-AUC values ranging from 0.64 to 0.67 in NLST and 0.60 to 0.65 in UIH cohorts across multiple years. Incorporation of imaging features to form a hybrid computer vision model significantly improved ROC-AUC values to 0.78-0.91 in NLST but deteriorated in UIH with ROC-AUC values of 0.68-0.80, attributable to Black participants where ROC-AUC values ranged from 0.63 to 0.72 across multiple years. Retraining the hybrid computer vision model by incorporating Black and other participants from the UIH cohort improved performance with ROC-AUC values of 0.70-0.87 in a held-out UIH test set. CONCLUSION: Hybrid computer vision predicted risk with improved accuracy compared with clinical risk models alone. However, potential biases in image training data reduced model generalizability in Black participants. Performance was improved upon retraining with a subset of the UIH cohort, suggesting that inclusive training and validation data sets can minimize racial disparities. Future studies incorporating vision models trained on representative data sets may demonstrate improved health equity upon clinical use.

  PublisherDOI

• NuGraph2 with context-aware inputs: physics-inspired improvements in semantic segmentation

  Journal of High Energy Physics · 2026-03-16

  articleOpen accessSenior author

  A bstract Graph neural networks have recently shown strong promise for event reconstruction tasks in Liquid Argon Time Projection Chambers, yet their performance remains limited for underrepresented classes of particles, such as Michel electrons. In this work, we investigate physics-informed strategies to improve semantic segmentation within the NuGraph2 architecture. We explore three complementary approaches: (i) enriching the input representation with context-aware features derived from detector geometry and track continuity, (ii) introducing auxiliary decoders to capture class-level correlations, and (iii) incorporating energy-based regularization terms motivated by Michel electron energy distributions. Experiments on MicroBooNE public datasets show that physics-inspired feature augmentation yields the largest gains, particularly boosting Michel electron precision and recall by disentangling overlapping latent space regions. In contrast, auxiliary decoders and energy-regularization terms provided limited improvements, partly due to the hit-level nature of NuGraph2, which lacks explicit particle- or event-level representations. Our findings highlight that embedding physics context directly into node-level inputs is more effective than imposing task-specific auxiliary losses, and suggest that future hierarchical architectures such as NuGraph3, with explicit particle- and event-level reasoning, will provide a more natural setting for advanced decoders and physics-based regularization. The code for this work is publicly available on Github at https://github.com/vitorgrizzi/nugraph_phys/tree/main_phys .

  PublisherOA PDFDOI

• PINN-FEM: A Hybrid Approach for Enforcing Dirichlet Boundary Conditions in Physics-Informed Neural Networks

  ArXiv.org · 2025-01-14

  preprintOpen accessSenior author

  Physics-Informed Neural Networks (PINNs) solve partial differential equations (PDEs) by embedding governing equations and boundary/initial conditions into the loss function. However, enforcing Dirichlet boundary conditions accurately remains challenging, often leading to soft enforcement that compromises convergence and reliability in complex domains. We propose a hybrid approach, PINN-FEM, which combines PINNs with finite element methods (FEM) to impose strong Dirichlet boundary conditions via domain decomposition. This method incorporates FEM-based representations near the boundary, ensuring exact enforcement without compromising convergence. Through six experiments of increasing complexity, PINN-FEM outperforms standard PINN models, showcasing superior accuracy and robustness. While distance functions and similar techniques have been proposed for boundary condition enforcement, they lack generality for real-world applications. PINN-FEM bridges this gap by leveraging FEM near boundaries, making it well-suited for industrial and scientific problems.

  PublisherOA PDFDOI

• Multi-Class Traffic Assignment using Multi-View Heterogeneous Graph Attention Networks

  ArXiv.org · 2025-01-15

  preprintOpen accessSenior author

  Solving traffic assignment problem for large networks is computationally challenging when conventional optimization-based methods are used. In our research, we develop an innovative surrogate model for a traffic assignment when multi-class vehicles are involved. We do so by employing heterogeneous graph neural networks which use a multiple-view graph attention mechanism tailored to different vehicle classes, along with additional links connecting origin-destination pairs. We also integrate the node-based flow conservation law into the loss function. As a result, our model adheres to flow conservation while delivering highly accurate predictions for link flows and utilization ratios. Through numerical experiments conducted on urban transportation networks, we demonstrate that our model surpasses traditional neural network approaches in convergence speed and predictive accuracy in both user equilibrium and system optimal versions of traffic assignment.

  PublisherOA PDFDOI

• Graph neural networks for travel distance estimation and route recommendation under probabilistic hazards

  International Journal of Transportation Science and Technology · 2025-02-26 · 14 citations

  articleOpen accessSenior author

  Estimating the shortest travel time and providing route recommendations between different locations in a city or region can quantitatively measure the conditions of the transportation network during or after extreme events. One common approach is to use Dijkstra’s Algorithm, which produces the shortest path as well as the shortest distance. However, this option is computationally expensive when applied to large-scale networks. This paper proposes a novel fast framework based on graph neural networks (GNNs) which approximate the single-source shortest distance between pairs of locations, and predict the single-source shortest path subsequently. We conduct multiple experiments on synthetic graphs of different sizes to demonstrate the feasibility and computational efficiency of the proposed model. In real-world case studies, we also applied the proposed method of flood risk analysis of coastal urban areas to calculate delays in evacuation to public shelters during hurricanes. The results indicate the accuracy and computational efficiency of the GNN model, and its potential for effective implementation in emergency planning and management.

  PublisherDOI

• Toward signed distance function based metamaterial design: Neural operator transformer for forward prediction and diffusion model for inverse design

  Computer Methods in Applied Mechanics and Engineering · 2025-08-18 · 8 citations

  articleOpen access

  The inverse design of metamaterial architectures presents a significant challenge, particularly for nonlinear mechanical properties involving large deformations, buckling, contact, and plasticity. Traditional methods, such as gradient-based optimization, and recent generative deep-learning approaches often rely on binary pixel-based representations, which introduce jagged edges that hinder finite element (FE) simulations and 3D printing. To overcome these challenges, we propose an inverse design framework that utilizes a signed distance function (SDF) representation combined with a conditional diffusion model. The SDF provides a smooth boundary representation, eliminating the need for post-processing and ensuring compatibility with FE simulations and manufacturing methods. A classifier-free guided diffusion model is trained to generate SDFs conditioned on target macroscopic stress-strain curves, enabling efficient one-shot design synthesis. To assess the mechanical response and the quality of the generated designs, we introduce a forward prediction model based on Neural Operator Transformers (NOT), which accurately predicts homogenized stress-strain curves and local solution fields for arbitrary geometries with irregular query meshes. This approach enables a closed-loop process for general metamaterial design, offering a pathway for the development of advanced functional materials.

  PublisherDOI

• Geometry-informed neural operator transformer for partial differential equations on arbitrary geometries

  Computer Methods in Applied Mechanics and Engineering · 2025-12-23 · 4 citations

  articleOpen access

  • A Geometry-Informed Neural Operator Transformer (GINOT) is proposed for forward predictions on arbitrary geometries. • GINOT encodes surface point clouds that are unordered, have non-uniform point density, and varying numbers of points. • GINOT effectively processes complex, arbitrary geometries and varying input conditions with good predictive accuracy. Machine-learning-based surrogate models offer significant computational efficiency and faster simulations compared to traditional numerical methods, especially for problems requiring repeated evaluations of partial differential equations. This work introduces the Geometry-Informed Neural Operator Transformer (GINOT), which integrates the transformer architecture with the neural operator framework to enable forward predictions on arbitrary geometries. GINOT employs a sampling and grouping strategy together with an attention mechanism to encode surface point clouds that are unordered, exhibit non-uniform point densities, and contain varying numbers of points for different geometries. The geometry information is seamlessly integrated with query points in the solution decoder through the attention mechanism. The performance of GINOT is validated on multiple challenging datasets, showcasing its accuracy and generalization capabilities for complex and arbitrary 2D and 3D geometries.

  PublisherDOI

Recent grants

• CAREER: Efficient Predictive Modeling for Infrastructure Systems Using Polynomial Approximation

  NSF · $500k · 2018–2024

Frequent coauthors

• Christopher W. Tessum

  University of Illinois Urbana-Champaign

  26 shared

• Sotiria Koloutsou‐Vakakis

  University of Illinois Urbana-Champaign

  26 shared

• Eleftheria Kontou

  University of Illinois Urbana-Champaign

  26 shared

• Lei Zhao

  25 shared

• Roger Ghanem

  14 shared

• Negin Alemazkoor

  12 shared

• Mohammad Amin Nabian

  Nvidia (United States)

  12 shared

• Weiheng Zhong

  10 shared

Labs

• Computational Intelligence for Engineering Lab (CIEL)PI

Education

• Ph.D., Computer Science

  University of Illinois at Urbana-Champaign

  2002

• M.S., Computer Science

  University of Illinois at Urbana-Champaign

  1998

• B.S., Computer Engineering

  University of Tehran

  1995

Awards & honors

• NSF CAREER Award on fast computational models for infrastruc…